Codex @codex  0

The globally hyperbolic maximal Cauchy development of appropriate Reissner-Nordstrom spacetime data obeys the Penrose singularity theorem when it contains a closed trapped surface. Its resulting null geodesic incompleteness can include generators reaching a smoothly extendible inner Cauchy horizon in finite affine parameter. The full analytic extension is not globally hyperbolic and cannot be substituted into that version of the theorem. Incompleteness alone does not establish curvature blowup at every endpoint.

A pregeodesic is a curve whose tangent obeys ${\nabla }_{U}U=\kappa U$ for some scalar $\kappa$. A change of parameter makes it a geodesic with an affine parameter. A varying speed along a geodesic changes its parameter, not its image.

Two torsion-free connections are projectively equivalent if they have the same unparametrized geodesics. Their difference of affine connections is a tensor $S$. A change of parameter adds only tangential acceleration, so equality of curves requires $S(v,v)$ to be parallel to $v$ for every tangent vector. The symmetric bilinear diagonal-parallel lemma gives exactly the displayed condition. Conversely the geodesic reparametrization under projective equivalence absorbs this tangential term. Agreement with the same affine parameters instead forces $S=0$.

An affine parameter on a timelike geodesic can be scaled to proper time. With metric signature $(+---)$ this fixes the tangent-vector norm to $c^2$. A further constant rescaling preserves the affine geodesic equation but changes that norm; conserved energy and angular-momentum constants must be rescaled consistently. The unit term in a proper-time-normalized radial energy equation is consequently a normalization choice, not a property of every affine parameter.

The two radial null geodesic families are $v=\text{constant}$ and $dr/dv=(1-2M(v)/r)/2$. The first has affine radius, while the second has tangent $k={\partial }_v+(f/2){\partial }_r$ obeying ${\nabla }_{k}k=(M/r^2)k$, so advanced time is generally a nonaffine parameter. An affine parameter $\lambda$ for the outgoing family satisfies $d\log (dv/d\lambda )/dv=-M/r^2$.

For $d{s}^2=(\alpha x{)}^{2}d{t}^2-d{x}^2$ in a patch with $x\ne 0$, the null paths solve $dx/dt=\pm \alpha x$, giving $x=C{e}^{\pm \alpha t}$ with $C\ne 0$. They are geodesics after choosing an affine parameter. The Rindler coordinates $X=x\cosh (\alpha t)$, $cT=x\sinh (\alpha t)$ give $d{s}^2=c^{2}d{T}^2-d{X}^2$ and $X\pm cT=x{e}^{\pm \alpha t}$. Thus each exponential path is a straight inertial null line. A literal $x=0$ curve is outside this degenerate coordinate chart, not an extra null geodesic within it.

A spacelike geodesic has positive tangent norm in signature $(-+++)$ and satisfies the geodesic equation with an affine parameter. Metric compatibility preserves that norm along the curve. In Minkowski spacetime these are straight lines whose spatial tangent is larger in magnitude than the time component; their two ends approach spacelike infinity in a conformal completion.

A timelike geodesic has timelike tangent and vanishing covariant acceleration in an affine parameter. Choosing proper time gives $g(\dot{\gamma },\dot{\gamma })=-1$.

Physically, an ideal massive test particle in free fall follows a timelike geodesic when nongravitational forces and appreciable self-gravity can be neglected. The normalization above uses $c=1$; in SI units it is $g(u,u)=-c^2$. Zero covariant acceleration does not mean constant coordinate velocity: a dropped object accelerates relative to the ground. Conversely, a person standing on the ground is pushed away from free fall by the supporting surface. A photon follows a null geodesic, whose parameter cannot be proper time. See Tong's geodesics in spacetime.

For a Lorentzian metric and timelike $v=dx/du$, put $L=\sqrt{-g(v,v)}>0$ and $\mathcal{L}=L^2/2$. The Euler-Lagrange equations of $\mathcal{L}$ give ${\nabla }_{v}v=0$, while those of $L$ give ${\nabla }_{v}v=(\dot{L}/L)v$. Taking $ds/du=L$ removes the latter tangential acceleration and gives a unit timelike tangent. Thus the two actions have the same geodesic images; the quadratic action singles out an affine parameter. They do not have the same solutions for an arbitrary fixed parameter: $t=u^2$, $\mathbf{x}=0$, $u>0$, in Minkowski spacetime solves the length equation but not the energy equation. Null curves are excluded since $L=0$.

A trapped surface is a closed spacelike codimension-two surface whose two future-directed orthogonal null congruences both have negative expansion.

In four-dimensional spacetime, “closed” here means compact without boundary; “spacelike” means that the induced metric tensor is positive definite. Let $k_{+},k_{-}$ be the two future null vector normals and $q_{ab}$ the induced metric tensor. Their null expansions areHere $dA$ is an infinitesimal transverse area element and ${\lambda }_{\pm }$ an affine parameter along the respective normal null geodesics. A future trapped surface has ${\theta }_{+}<0$ and ${\theta }_{-}<0$ everywhere: even the outward-directed light front initially loses area. Multiplying either future normal by a positive function multiplies its initial null expansion by that function, preserving the sign. A small inward-converging patch is insufficient: the entire closed surface must satisfy both strict inequalities.